Effect Hyperglycemia Vivo Secretion Partially ... · glucose profile by phlorizin treatment in diabetic rats com-pletely corrected all three beta cell abnormalities. These re-

Effect of Chronic Hyperglycemia on In Vivo Insulin Secretion inPartially Pancreatectomized RatsLuciano Rossetti, Gerald 1. Shulman, Walter Zawalich, and Ralph A. DeFronzoYale University School of Medicine and School of Nursing, NewHaven, Connecticut 06510

Abstract

Wehave examined the effect of chronic (4 wk) hyperglycemiaon insulin secretion in vivo in an awake, unstressed rat model.Three groups of animals were examined: control, partial (90%)pancreatectomy, and partial pancreatectomy plus phlorizin, inorder to normalize plasma glucose levels. Insulin secretion inresponse to arginine (2 mM), hyperglycemia (+100 mg/dl),and arginine plus hyperglycemia was evaluated. In diabeticcompared with control animals three specific alterations wereobserved: (a) a deficient insulin response, in both first andsecond phases, to hyperglycemia; (b) an augmented insulin re-sponse to the potentiating effect of arginine under basal glyce-mic conditions; and (c) an inability of hyperglycemia to aug-ment the potentiating effect of arginine above that observedunder basal glycemic conditions. Normalization of the plasmaglucose profile by phlorizin treatment in diabetic rats com-pletely corrected all three beta cell abnormalities. These re-sults indicate that chronic hyperglycemia can lead to a defect inin vivo insulin secretion which is reversible when normoglyce-mia is restored.

Introduction

Glucose is the primary regulator of insulin secretion in vivo. Itexerts this role by stimulating directly the beta cell and bymodulating the effect of other secretagogues (1, 2). In contrastto acute intermittant elevations in the plasma glucose concen-tration, which provide a powerful insulin secretory stimulus tothe beta cell, Weir et al. have advanced the hypothesis thatchronic sustained hyperglycemia, in the presence of a reducedbeta cell mass, may lead to the development of a defect ininsulin secretion (3). Over the years, much evidence has accu-mulated to indicate that the beta cells of diabetic man (4, 5)and animals (6-9) have acquired a specific defect in their abil-ity to respond to glucose. In contrast, the pancreas' response tononglucose secretagogues such as arginine (9), leucine (7), sul-fonylurea drugs (10), and isoproterenol (10, 11) is well main-tained. Similarly, the ability of glucose to modulate the insulinresponse to these nonglucose secretagogues is consistently im-paired (7, 9). However, the precise factors that are responsiblefor the beta cells' "blindness" to glucose have yet to be defined.

Address correspondence to Dr. Rossetti, 2071 LMPBldg., Yale-NewHaven Hospital, 333 Cedar St., NewHaven, CT 06510.

Received for publication 23 January 1987 and in revised form 18May 1987.

In the present study we have examined, for the first time invivo, whether chronic hyperglycemia can lead to the develop-ment of a defect in insulin secretion in otherwise healthy betacells, and whether restoration of normoglycemia can correctthe impairment in insulin secretion. To bring about a reduc-tion in beta cell mass and produce a state of chronic hypergly-cemia, a 90% surgical pancreatectomy was performed (12).The remaining 10% of the pancreas is left intact, and insulinsecretion, expressed per remaining pancreatic mass, should benormal unless some acquired defect in beta cell function de-velops. To examine whether any acquired beta cell abnormal-ity was specifically the result of chronic exposure to hypergly-cemia, diabetic rats received a continuous infusion of phlori-zin for 4 wk. This agent induces renal glucosuria and leads tonormal fasting and near normal postmeal plasma glucoselevels (13). The ability of the pancreas to secrete insulin wasevaluated using the hyperglycemic clamp technique (14) sothat the early and late phases of insulin secretion could beexamined separately and compared with the response to argi-nine.

Methods

Animal preparation. Three groups of male Sprague-Dawley rats(Charles River Breeding Laboratories, Wilmington, MA), weighingbetween 80 and 100 g (3-4 wk old) were studied. Group I: sham-oper-ated controls (n = 7); group II: partially pancreatectomized rats (n= 10); and group III: partially pancreatectomized rats treated withphlorizin (n = 6). All rats (3-4 wk old) were anesthesized with pheno-barbital (50 mg/kg body wt), and in groups II and III 90% of thepancreas was removed according to the technique of Foglia (12), asmodified by Bonner-Weir et al. (3). Group I animals underwent asham pancreatectomy in which the pancreas was disengaged from themesentery and gently rubbed between the fingers. Postoperatively, ratswere housed in an environmentally controlled room with a 12-h light/dark cycle and allowed free access to standard rat chow (Ralston-Pur-ina Co., St. Louis, MO)and water. Within 48 h all animals had recov-ered from surgery and were eating normally. In group III rats phlorizin(0.4 g/kg body wt per d made up in a 20%solution of propylene glycol)treatment was initiated 14 d postsurgery, at a time when hyperglyce-mia first became evident in the partially pancreatectomized rats andwas continued for 4 wk. Phlorizin was administered as a continuoussubcutaneous infusion through a small implantable minipump (Alzetosmotic minipump; Alza Corp., Palo Alto, CA) to ensure day-longinhibition of renal tubular glucose reabsorption. After surgery, ratswere weighed twice weekly and tail vein blood collected at the sametime for the determination of fed plasma glucose and insulin concen-trations.

5 wk postpancreatectomy and sham operation (i.e., I wk beforeperforming the hyperglycemic clamp study) rats were anesthesizedwith phenobarbital (50 mg/kg body wt) and indwelling catheters wereinserted so that the animals could be studied in the awake, unstressedstate. Two internal jugular catheters were inserted and extended to thelevel of the right atrium and a left carotid catheter was advanced to theaortic arch. The three catheters were filled with heparnn/polyvinylpyr-

Effect of Chronic Hyperglycemia on In Vivo Insulin Secretion 1037

J. Clin. Invest.© The American Society for Clinical Investigation, Inc.0021-9738/87/10/1037/08 $2.00Volume 80, October 1987, 1037-1044

rolidone solution, sealed, and tunneled subcutaneously around theside of the neck to the back of the head. The catheters were externalizedthrough a skin incision and anchored to the skull with a dental cementcap (Durelon; Premier Co., Norristown, PA), which was reinforced byfour jeweler's case screws (0.07 mm).

Experimental protocol. All studies were conducted in the morningafter a 10-12-h overnight fast. Throughout the study, the rats wereallowed to move freely within the confines of a large cage and theconnecting tubing was suspended overhead by means of a pulley sys-tem. The venous catheter was used for blood withdrawal and the arte-rial catheter for infusion of all test substances. To prevent intravascularvolume depletion and anemia, fresh whole blood obtained by heartpuncture from littermates of the test animal was administered at aconstant rate designed to quantitatively replace the total blood lossduring the study.

Arginine/hyperglycemic clamp study (Fig. 1). Blood was obtainedfor the determination of plasma glucose and insulin concentration at-30 and 0 min. At time zero a prime-continuous infusion of argininesolution (300 mg/ml) was administered from time 0-30 min to acutelyraise and maintain the plasma atginine concentration at - 2 mM.Theprime was administered at the rate of 280 Amol/min from 0 to 2 minand the continuous infusion was administered at 36 Mmol/min from 2to 30 min. Blood for the determination of plasma glucose and insulinconcentrations was obtained at time 2, 4, 6, 8, 10, 20, and 30 min afterstarting the arginine infusion. Between time 30 and 80 min a constantsaline infusion (15 Ml/min) was administered to maintain the arterialcatheter patent. Blood samples were withdrawn at time 60 and 80 minfor determination of plasma glucose and insulin levels. At time 80 mina priming infusion ofd25% glucose was administered to acutely raise theplasma glucose concentration by - 100 mg/dl above baseline. Theplasma glucose concentration was subsequently held constant at thishyperglycemic plateau until time 170 min by the adjustment of avariable glucose infusion based upon a negative feedback principle(14). Plasma samples for determination of glucose and insulin wereobtained at 2-min intervals between time 80 and 90 min and at 10-minintervals thereafter until 140 min. From 140 to 170 min a prime-con-tinuous infusion of arginine was again administered as previously de-scribed. Plasma arginine concentration was determined at time -30, 0,10, 30, 80, 150, and 170 min.

Pancreatic weight and insulin content. 4 d after the arginine/hy-perglycemic clamp study all rats, after a 12-h overnight fast, wereanesthesized and the pancreas was quickly dissected, separated fromgut and major lymph nodes, rinsed in saline, blotted, and weighed. Thepancreatic tissue was then frozen in liquid nitrogen and stored at-30'C for subsequent insulin analysis. The tissue was homogenized inice-cold acid-ethanol (0.15 MHCI-75% ethanol) and incubated for 48h at 4VC. After incubation the suspension was centrifuged for 10 min at10,000 rpm in a Sorval refrigerated centrifuge. The supernatant wasdecanted and a 1:1,000 dilution was prepared in phosphate-bufferedsaline solution containing 0.25% bovine serum albumin. Insulin con-centration was measured in quadruplicate using a rat insulin standard

(Eli Lilly & Co., Indianapolis, IN). All insulin determinations wereperformed in the same assay to eliminate any interassay variations.

Analytical procedures. Plasma glucose was measured by the glucoseoxidase method (glucose analyzer, Beckman Instruments Inc., PaloAlto, CA) and plasma insulin by radioimmunoassay using rat insulin(lot 615-163-12-3; Eli Lilly & Co.) as standard (15). Plasma arginineconcentration was determined by an automated ion exchange chro-matographic technique (D-500; Dionex Corp., Sunnyvale, CA) with anL6 column and lithium citrate buffer.

Calculations. The plasma insulin responses presented in TablesIII-V represent the mean increment in plasma insulin concentrationabove baseline during the 0-30 (arginine infusion)-, 80-140 (glucoseinfusion)-, 140-170 (arginine plus glucose infusion)-min time inter-vals. The basal insulin concentration for the arginine infusion wascalculated as the mean of the -30- and 0-min plasma insulin concen-trations. The baseline plasma insulin concentration for the glucoseinfusion was calculated as the mean of the 70- and 80-min time points.The baseline plasma insulin concentration for the combined glucose/arginihe infusion period was calculated as the mean of the 130- and140-min time points. All values are presented as the meAn±SEM.Significant differences between groups were determined using theone-way analysis of variance in conjunction with the Student-New-man-Kuel test.

Results

Body weight, plasma glucose, and insulin concentrations. Allanimals were studied 6 wk after pancreatectomy or sham pan-createctomy. At this time no significant differences in bodyweight were observed between any of the three groups (TableI). The fasting (mean of four values during the 2 wk beforestudy) and fed plasma glucose concentrations are shown inTable I. In group II rats (90% pancreatectomy) the fastingplasma glucose concentration was slightly increased (both P<0.01) compared with both groups I (sham-operated controls)and III (phlorizin-treated diabetic rats). The fasting plasmaglucose in phlorizin-treated diabetic animals was slightly butnot significantly elevated compared with controls. The fedplasma glucose concentration in diabetic rats was approxi-mately twice that observed in controls (P < 0.001). Before theinitiation of phlorizin treatment at 2 wk after pancreatectomy,the fed plasma glucose (296±19 mg/dl) was similar to thatobserved in the diabetic group (284±13 mg/dl). After 4 wk ofphlorizin therapy (group III), the fed plasma glucose concen-tration was reduced to levels similar to those in controls.

The fasting plasma insulin concentration was similar in allthree groups. The fed plasma insulin concentration was simi-larly reduced in groups II and III compared with controls (P<0.001).

Table I. Fasting and Fed Plasrha Glucose and Insulin Concentrations in Control and Diabetic Rats*

Fasting plasma Fasting plasmaGroup Number Weight glueose Fed plasma glucose insulin Fed plasma insulin

g mg/dl mg/dl ng/ml ng/mlI Sham operated 7 276±7 103±2 143±2 2.44±0.16 5.79±0.24

II Diabetic 10 264±9 119±3* 284±13§ 2.37±0.20 3.04±0.151III Phlorizin-treated diabetic 6 258±11 111±3 142±6 2.50±0.15 2.95±0.20i

(296±19)"

* All values represent the mean±SEM. $ P < 0.01 vs. sham-operated control rats. 1P < 0.001 vs. sham-operated and phlorizin-treated dia-betic rats. "l Fed plasma glucose concentration in the same six rats before the institution of phlorizin treatment (i.e., at week two).

1038 L. Rossetti, G. I. Shulman, W. Zawalich, and R. A. DeFronzo

Pancreatic insulin content (Table II). Pancreatic weightand insulin content in groups II and III were similarly andmarkedly decreased to - 15% of the sham-operated controlgroup (P < 0.001). The decrease in pancreatic insulin contentclosely paralleled the reduction in pancreatic mass in all rats.

Plasma glucose and insulin concentrations during argi-nine/glucose infusion (Figs. I and 2; Table III). After an over-night fast (2200-0900 h) plasma arginine concentrations were152±4 ,M in group I, 194±12 MAMin group II, and 187±15MMin group III. Arginine infusion (0-30 min) resulted insteady state plasma arginine levels that were similar in all threegroups: 2,187±199 M&Min group I, 2,319±99 MuMin group II,and 2,357±107 AMin group III. The plasma glucose concen-tration increased slightly (by 10-20 mg/dl) and similarly in thethree groups during arginine infusion (Fig. 1). The mean in-cremental plasma insulin response to arginine was reduced by54% in the diabetic group and by 84% in the diabetic ratstreated with phlorizin (P < 0.00 1) (Table III).

By the end of the 30-80-min period of saline infusion,plasma glucose and insulin concentrations returned to basal,pre-arginine infusion levels (Figs. 1 and 2). During the 1 hhyperglycemic clamp study (80-140 min) the increment inplasma glucose concentration was 96±2, 99±4, and 99±4mg/dl in groups I, II, and III, respectively. In all studies thecoefficient of variation in plasma glucose concentration was< 5%. The glucose infusion rate (140-180-min period) neces-sary to maintain the plasma glucose concentration constant atthe desired hyperglycemic plateau was 24.3+1.2, 4.3+0.3, and7.3±0.4 mg/kg. min in groups I, II, and III, respectively. Themean plasma insulin response in the diabetic group was re-duced by 95% compared with controls (0.30 vs. 5.75 ng/ml, P< 0.001). In phlorizin-treated diabetic rats the mean plasmainsulin response (0.91 ng/ml) was increased threefold (P< 0.01) compared with the diabetic group, but was still muchreduced compared with controls when viewed in absoluteterms (Table III).

When an arginine infusion (140-170 min) was superim-posed on hyperglycemia, the mean incremental plasma insulinresponse (20.61 ng/ml) was significantly greater than the addi-tive effects of hyperglycemia (5.75 ng/ml) alone or arginine(3.80 ng/ml) alone. During combined arginine/glucose infu-sion, the mean plasma incremental response in both group IIand group III rats was reduced by 90% compared with con-trols (P < 0.001).

Since diabetic animals had 90%of their pancreas removed,it is not surprising that the plasma insulin response, whenviewed in absolute terms, was deficient. Therefore, the incre-mental plasma insulin responses were also calculated per gramof pancreatic tissue and per pancreatic insulin content (TableIV). Since pancreatic weight and insulin content were highly

Table II. Pancreatic Weight and Insulin Contentin Control and Diabetic Rats

Group Number Pancreatic weight Insulin content

g ug/pancreas

I Shamoperated 7 1.32±0.031 51.7±12.4II Diabetic 10 0.194±0.012 8.5±3.2

III Phiorizin-treated diabetic 6 0.188±0.020 7.1±2.8

ARGININE

240

220

200E

C 180

> 160-a

Q 1401-

120

100

ARGNINEI

,&SHAM OPERATED* DIABETICo PHORIZIN-TREATED

DIABETIC

-1

-30 0 20 30 80 90 100 110 120 130 140 150 160 170

TIME (min)

Figure 1. Plasma glucose concentration (milligrams per milliliter)during arginine infusion (t = 0-30 min), during the hyperglycemicclamp (t = 80-140 min), and during combined arginine plus glucoseinfusion (t = 140-170 min) in group I (sham operated, open trian-gles), in group II (diabetic, solid circles), and in group III (phlorizin-treated diabetic, open squares).

correlated (r = 0.95, P < 0.00 1), both methods of data expres-sion yielded qualitatively and quantitatively similar results.The incremental plasma insulin response to arginine in dia-

ARGININE ARGININEGLUCOSE

40

= 30

z

-J

=, 20

0. 10-2

8

6

E0.

C

.E-jCl):m

X.U)ct

-iCL

4

2

a SHAM OPERATED

I

4i4 s -_ t._..+__ .__

iiPiI

'i

v 4

i f---'1f,. ..P-30 0 10 20 30 80 90 100 110 120 130 140 150 160 170

TIME (min)

* DIABETIC* PHORIZIN-TREATED

DIABETIC

hi' Ad= 'I -

44,,-30 0 10 20 30 o 90 100 110 120 130 140 Iso 160 170

TIME (min)

Figure 2. Plasma insulin concentration (nanograms per milliliter)during arginine infusion (t = 0-30 min), during the hyperglycemicclamp (t = 80-140 min), and during combined arginine plus glucoseinfusion (t = 140-170 min) in group I (sham operated, open trian-gles, top), group II (diabetic, solid circles, bottom), group III (phlori-zin-treated diabetic, open squares, bottom).


Table III. Plasma Glucose Concentrations and Mean Incremental Plasma Insulin Responses after Arginine,Glucose, and Combined Arginine/Glucose Infusion in Control and Diabetic Rats

Mean plasma glucose concentration (mg/dl) Mean increment plasma insulin concentration (ng/ml)

Arginine ArginineArginine Hyperglycemia + hyperglycemia Arginine Hyperglycemia + hyperglycemia

Group (0-30 min) (80-140 min) (140-170 min) (0-30 min) (80-140 min) (140-170 min)

I Sham operated 115±3 211±2 213±4 3.80±0.41 5.75±0.75 20.61±3.26II Diabetic 122±5 221±4 225±5 1.73±0.35* 0.30±0.05* 2.46±0.62*

III Phlorizin-treated diabetic 126±3 225±5 223±5 0.52±0.15* 0.91±0.09*t 2.21±0.53*

* P < 0.001 vs. group I. t P < 0.01 vs. group II.

betic rats was increased threefold compared with controls (P< 0.001), and phlorizin treatment normalized this hyperinsu-linemic response (Table IV). In contrast to arginine, the incre-mental insulin response to hyperglycemia was reduced by

- 65% in the diabetic group (1.5 vs. 4.4 ng/ml per g pancreas),and this defect in insulin secretion was totally corrected byphlorizin treatment (4.8 ng/ml per g pancreas). Whenarginineinfusion was superimposed upon hyperglycemia (140-170-min time period), the incremental plasma insulin response wasslightly, although not significantly, reduced in both diabetic(12.7 ng/ml per g pancreas) and phlorizin-treated diabetic(11.7) groups compared with controls (15.6).

The modulating effect of glucose on arginine-induced in-sulin secretion was calculated as the difference between theincremental insulin response to the second arginine stimulusminus that induced by the first arginine stimulus. The abilityof hyperglycemia to augment arginine-stimulated insulin se-cretion was clearly impaired in diabetic compared with controlrats (3.01±1.06 vs. 12.46±1.69 ng/ml per g pancreas, P< 0.001). Phlorizin treatment increased the modulating effectof hyperglycemia on arginine-induced insulin secretion to avalue (9.09±1.55 ng/ml per g pancreas) that was not signifi-cantly different from controls.

First and second phases of insulin secretion (Table V). Thefirst phase of insulin secretion (calculated as the mean incre-mental plasma insulin concentration during the initial 10 min,i.e., 80-90-min time period of the hyperglycemic clamp) wasreduced by 99% in diabetic rats, while the second phase ofinsulin secretion (calculated as the mean incremental plasmainsulin concentration during the 90-140-min time period ofthe hyperglycemic clamp) was diminished similarly by 94%

(Table V, top). Phlorizin treatment resulted in significant im-provements in both the first and second phases of insulin se-cretion. The effect of phlorizin to augment the first and secondphases of insulin secretion is more dramatically demonstratedwhen the incremental plasma insulin response is expressed pergram of pancreatic weight (Table V, bottom). As can be seen,both the first (4.36 vs. 3.67 ng/ml per g pancreas) and second(4.89 vs. 4.50 ng/ml per g pancreas) phase insulin responseswere completely normalized in the phlorizin-treated diabeticrats.

Discussion

At substrate concentrations within the physiologic range, glu-cose is the primary regulator of insulin secretion (16). It notonly stimulates insulin release directly, but also modulates theaction of other secretagogues on the beta cell ( 17). In normalman chronic hyperglycemia has been shown to augment invivo insulin secretion (18, 19). In contrast, in rats with a re-duced beta cell mass evidence has accumulated that chronichyperglycemia may lead to a defect in beta cell function bydecreasing glucose-stimulated insulin secretion and impairingthe modulating effect of nonglucose secretagogues on insulinsecretion (6-9, 20). Note, however, that all of the human stud-ies involved healthy, nondiabetic subjects and the period ofhyperglycemia was of short duration (days), whereas the ani-mal studies involved prolonged periods of hyperglycemia(weeks) and were carried out in diabetic rats. Furthermore, allof these previous studies have two major shortcomings: First,islet cell function was examined in vitro and, to our knowl-

Table IV. Mean Plasma Insulin Response Expressed per Gram Pancreatic Weight or per UnitPancreatic Insulin Content in Control and Diabetic Rats

Mean plasma insulin concentration per pancreatic weight Mean plasma insulin concentration per pancreatic insulin(ng/ml/g pancreas) content (ng/ml/ng pancreatic insulin)

Arginine ArginineGroups Arginine Hyperglycemia + hyperglycemia Arginine Hyperglycemia + hyperglycemia

I Sham operated 2.9±0.2 4.4±0.5 15.6±2.4 73±7 111±14 399±63II Diabetic 8.9±1.7* 1.5±0.3* 12.7±3.2 204±40* 35±6* 289±73

III Phlorizin-treated diabetic 2.8±0.7 4.8±0.5 11.7±2.8 73±20 128±13 311±75

* P < 0.01 vs. sham-operated control and phlorizin-treated diabetic rats.


Table V. First and Second Phase Plasma InsulinResponses during the Hyperglycemic Clamp StudyPerformed in Control and Diabetic Rats

Plasma insulin response (ng/ml/min)

First phase Second phaseGroup (80-90 min) (90-140 min)

I Shamoperated 4.84±0.53 5.94±0.86II Diabetics 0.05±0.1 1* 0.36±0.06*t

III Phlorizin-treated diabetics 0.82±0.08* 0.92±0.11*

Plasma insulin response(ng/mI/min/g pancreas)

First phase Second phaseGroup (80-90 min) (90-140 min)

I Shamoperated 3.67±0.37 4.50±0.65II Diabetics 0.25±0.57* 1.86±0.31*

III Phlorizin-treated diabetics 4.36±0.37 4.89±0.58

* Different from group I, P < 0.001.Different from group III, P < 0.001.

edge, the impairment of beta cell function after chronic hyper-glycemia has not been characterized in vivo. This is of consid-erable importance since the neuronal input to the pancreasand the influence of circulating hormones are lost when theislets are removed from the body. Moreover the precedinganesthesia, with an increase in circulating catecholaminelevels, may alter beta cell function. Second, and more impor-tantly, before the concept that chronic hyperglycemia is delete-rious to beta cell function can be accepted, it must be shownthat restoration of normoglycemia can return insulin secretionto normal. In the one rat study in which insulin treatment wasused to normalize plasma glucose levels, no improvement inglucose-induced insulin secretion was observed (9). Interest-ingly, short-term insulin therapy in man has been shown toresult in a partial improvement in insulin secretion in type IIdiabetic individuals (21, 22). However, it is not clear whetherthis improvement is related to the decline in plasma glucoseconcentration or to some other metabolic effect brought aboutby insulinization.

In the present study we have examined the effect of chronichyperglycemia on insulin secretion in an awake, unstressed ratmodel. Partial (90%) pancreatectomy was used to reduce betacell mass and cause a chronic sustained, modest elevation inthe plasma glucose concentration. With this procedure theresidual 10% of the pancreas lying between the common bileduct and the duodenum is not touched, so that the remainingbeta cells should be functionally intact. This avoids the con-cern of beta cell toxicity when streptozotocin is used to reducebeta cell mass. To evaluate beta cell function, we used thehyperglycemic clamp technique (14) since this allows the in-vestigator to examine both the first and second phases of insu-lin secretion. The loss of first phase insulin secretion has beenpostulated to represent a specific and early defect in the patho-genesis of both type I (23, 24) and type II (24-26) diabetesmellitus. Furthermore, its reversal with treatment of the hyper-glycemia has never been reported in animal models of dia-betes; in man, only occasional reports have demonstrated a

partial correction after normalization of the plasma glucoseconcentration (27, 28).

6 wk after partial pancreatectomy the fasting plasma glu-cose concentration had risen to - 120 mg/dl, and moderatelysevere postmeal hyperglycemia (250-300 mg/dl) was present.These fed glucose values are similar to those reported by Or-land et al. (29) and slightly greater than those reported byBonner-Wier et al. (3). These small differences in the severityof diabetes is most likely accounted for by a number of factorsincluding strain differences, variations in the age at which thesurgical pancreatectomy was performed, and interanimal vari-ation (30). Fasting plasma insulin levels were similar in dia-betic and control rats but the post-meal plasma insulin re-sponse (Table I) was markedly impaired. When diabetic ani-mals were treated with phlorizin, an inhibitor of renal glucosetransport, both the fasting and post-meal plasma glucose con-centrations were returned to normal without any change ineither the basal or fed plasma insulin levels. Since phlorizintreatment did not affect pancreatic weight or pancreatic insu-lin content, any improvement in beta cell function can not beattributed to an increased beta cell mass or insulin content.Pancreatic weight and insulin content was similar in diabeticand phlorizin-treated diabetic rats and was - 15% of that ob-served in the control animals. These data are in good agree-ment with those of Orland et al. (29), who showed that at 3 and14 wk after pancreatectomy the remnant pancreatic weightand insulin content ranged from 12 to 15% of normal.

To examine the modulating effect of glucose on nonglu-cose secretagogues, we quantitated arginine-mediated insulinsecretion at two plasma glucose concentrations (120 and 220mg/dl) in control, diabetic, and phlorizin-treated diabetic rats.Since 90% of the pancreas was removed in both diabeticgroups, it is obvious that the absolute amount of insulin se-creted must be reduced. Therefore, the plasma insulin re-sponse was expressed, not only in absolute terms, but also perresidual pancreatic weight and insulin content. During argi-nine infusion the basal plasma glucose concentration was simi-lar in diabetics compared with control rats. The absence of anexcessive glucose response to arginine often observed in otherdiabetic states can probably be related to the preserved ratio ofbeta/alpha cells in the pancreatectomized rat model (3). Simi-lar steady state plasma arginine levels were achieved in control,diabetic, and phlorizin-treated diabetic rats. Diabetic animals,whose beta cells had been exposed to day-long hyperglycemia,demonstrated a marked hyperresponsiveness (threefoldgreater) to arginine when studied at their fasting plasma glu-cose concentration (- 120 mg/dl). These results are qualita-tively quite similar to those of Leahy et al. (9), who used per-fused pancreata from partially pancreatectomized and strep-tozotocin diabetic rats. Thus, both the in vitro studies, as wellas our in vivo studies, demonstrate that chronic hyperglycemiahas a potentiating effect on amino acid-mediated insulin se-cretion.

In marked contrast to the potentiating effect on arginine-mediated insulin release, chronic hyperglycemia severely im-paired by 60-70% (Table IV) the pancreatic response to anacute glucose infusion. This reduction affected both the early(0-10 min) and late (10-60 min) plasma insulin responses,which were decreased by 93 and 60%, respectively. This severereduction in glucose-induced insulin secretion in vivo is con-sistent with a number of in vitro studies carried out with islet


tissue obtained from diabetic rats (6-9). Most recently, it hasbeen shown that a 48-h glucose infusion, designed to raise theplasma glucose concentration to 183±13 mg/dl in normal rats,leads to a defect in the beta cells' ability to respond to an acutehyperglycemic challenge in vitro (31). However, in the samestudy when glucose was infused for a similar period of time toraise the plasma glucose concentration to 164±4 or 370±25mg/dl, no impairment in the insulin secretory response toacute hyperglycemia was observed. To ensure that the inabilityof our diabetic rats to respond during the hyperglycemic clampwas not due to beta cell exhaustion, animals were rechallengedwith arginine while maintaining hyperglycemia at 220 mg/dl.As can be seen in Table IV and Fig. 2 (bottom), the insulinresponse to this second arginine stimulus was slightly, al-though not significantly less than observed in controls, if ex-pressed per pancreatic weight or insulin content. Thus, thedeficient response to hyperglycemia can not be attributed tobeta cell insulin depletion. Note that the incremental plasmainsulin response to the second arginine stimulus in diabeticrats was of similar magnitude to the initial arginine challenge,indicating that the potentiating effect of glucose was alreadynear maximal at a glucose concentration (120 mg/dl) similarto the fasting glucose level observed in the diabetic rats in vivo.This situation is quite different from that observed in the con-trol animals where the second arginine stimulus (at a plasmaglucose concentration of 220 mg/dl) produced a fivefoldgreater insulin response (Table IV) than observed during thefirst arginine stimulus. Thus, the diabetic beta cell appears tobe characterized by three specific alterations: (a) a deficientinsulin response, both first and second phases, to hyperglyce-mia; (b) a hypersensitivity to the potentiating effect of arginineunder basal glycemic conditions; and (c) an inability of hyper-glycemia to augment the potentiating effect of arginine abovethat observed under basal glycemic conditions. Our interpre-tation of the arginine infusion studies is that the prolongedhyperglycemia observed in diabetic rats induces a state ofchronic potentiation which is already maximal under the fast-ing hyperglycemic conditions observed in vivo.

To examine the role of hyperglycemia per se on the abovethree defects in insulin secretion, a separate group of diabeticanimals was started on phlorizin treatment 2 wk postpancre-atectomy, i.e., at the time when hyperglycemia is first noted todevelop. Phlorizin, by inhibiting glucose reabsorption by therenal tubular cells, induced a state of persistent renal gluco-suria and completely normalized both the fasting and post-meal plasma glucose levels without any alteration in theplasma insulin response. As can be seen in Table IV and Fig. 2(top), normalization of the plasma glucose profile corrected allof the defects in insulin secretion observed in diabetic rats.Most noteworthy, the plasma insulin response to hyperglyce-mia by the residual beta cells from diabetic rats was completelynormalized. This improvement in insulin secretion involvedboth the first and second phases of insulin release, which re-turned to values similar to those observed in control rats(Table V). One previous investigation has attempted to exam-ine whether glycemic control in diabetic (streptozotocin in-duced) rats could reverse the defect in glucose-stimulated in-sulin secretion (22); in this study insulin therapy failed to im-prove beta cell function. However, there were several noteabledifferences between this study (22) and our own. First, the

period of insulin therapy was quite short, only 24 h, and thismay have been of insufficient time to observe any improve-ment in insulin secretion. Second, the insulin treatment re-sulted in modest hypoglycemia, which of itself could impairinsulin secretion. And lastly, exogenous insulin administrationwith resultant hyperinsulinemia has been shown to have afeedback inhibition of pancreatic insulin secretion (32). In thestudies of Leahy et al. (31), where chronic (48 h) glucose infu-sion was associated with an acquired insulin secretory defect inresponse to an acute hyperglycemic stimulus, cessation of theglucose infusion for 72 h was associated with a return of insu-lin secretion to normal.

Long-term correction of hyperglycemia in diabetic rats alsocorrected the hyperresponsiveness to arginine. Thus, when theplasma arginine concentration was increased to 2 mMandplasma glucose was maintained at basal levels, the meanplasma insulin response decreased from 8.9 to 2.8 ng/ml per gpancreas (Table IV), a value that was not significantly differentfrom the control rats. Similarly, the potentiating effect of hy-perglycemia (220 mg/dl) on arginine-mediated insulin secre-tion increased markedly to a value that was slightly, althoughnot significantly, greater than in control animals.

Our observations may have particular relevance for under-standing the pathogenesis of type II or non-insulin-dependentdiabetes mellitus. The little data that are available in mansuggest that pancreatic mass is reduced to approximately60-70% of normal (33, 34) in non-insulin-dependent diabetesmellitus. Such a decrease, of itself, would not be sufficient toaccount for the marked insulinopenia observed in many typeII diabetic patients, especially in those with fasting plasmaglucose levels above 200-220 mg/dl. It seems likely, therefore,that some factor(s) must be responsible for impairing insulinsecretion in such individuals. On the basis of our results it isinteresting to hypothesize that one such factor could be glucoseitself. Thus, chronic hyperglycemia could be postulated tocause a "desensitization" of the beta cells' ability to respond toa glucose stimulus. Such a desensitization may be commontoall cells in the body. Using the same partially pancreatecto-mized rat model, we recently have shown that chronic hyper-glycemia leads to peripheral tissue insensitivity to the action ofinsulin (13). Other investigators have shown that hyperglyce-mia can down-regulate the glucose transport system in non-insulin dependent tissues, such as the cerebral cortex (35, 36).One could speculate that chronic sustained hyperglycemiacauses a generalized down-regulation of glucose transport andmetabolism in all cells of the body. This could account forsuch diverse observations as severe insulin resistance in pe-ripheral (13, 37) and hepatic (38) tissues, impaired insulinsecretion in response to glucose (but a normal response tononglucose stimuli) (6-9, 39), and the development of symp-toms of neuroglycopenia at "normal" plasma glucose levels inpoorly controlled diabetic individuals (40). Our results alsomight help to explain the uniform improvement in insulinsecretion observed after a number of diverse maneuvers, all ofwhich have in commonthe ability to lower the plasma glucoseconcentration. Thus, acute (41) and chronic (42) caloric re-striction, intensified insulin therapy (43), and sulfonylureadrugs (44, 45) all result in a significant lowering of the plasmaglucose concentration and a concomitant increase in insulinsecretion which may be sustained even after the therapy is

1042 L. Rossetti, G. L Shulman, W. Zawalich, and R. A. DeFronzo

discontinued. The therapeutic efficacy of these maneuversmay, in part, be related to their plasma glucose lowering effect,which secondarily leads to an improvement in insulin secre-tion.

In summary, our results demonstrate two important rela-tionships between the plasma glucose concentration and insu-lin secretion: first, chronic (4 wk) sustained hyperglycemiamarkedly impairs the ability of the beta cell to respond to anacute glucose stimulus; second, chronic hyperglycemia in-duces a state of beta cell "potentiation" to nonglucose secreta-gogues such as arginine. These alterations in insulin secretioncan be reversed completely with correction of the hyper-glycemia.

Acknowledgments

The authors would like to thank Miss Jo Anne Palmieri for her expertsecretarial assistance. Ms. Vicky Diaz provided technical assistance inperforming the insulin determinations.

This work was supported in part by a grant from the JuvenileDiabetes Foundation (JDF185279 to G. I. Shulman). Dr. Rossetti isthe recipient of JDF postdoctoral fellowship 386266. Weare thankfulto Drs. Susan Bonner-Wier and Gordon Wier for teaching us how toperform the partial pancreatectomy technique.

References

1. Gerich, J., M. Charles, and G. Grodsky. 1974. Characterizationof the effects of arginine and glucose on glucagon and insulin releasefrom the perfused rat pancreas. J. Clin. Invest. 54:833-842.

2. Pagliara, A., S. Stillings, B. Hover, D. Martin, and F. Mats-chinsky. 1974. Glucose modulation of amino acid-induced glucagonand insulin release in the isolated perfused rat pancreas. J. Clin. Invest.54:819-832.

3. Bonner-Weir, S., D. F. Trent, and G. C. Weir. 1983. Partialpancreatectomy in the rat and subsequent defect in glucose-inducedinsulin release. J. Clin. Invest. 71:1544-1553.

4. Robertson, R. P., and D. Porte, Jr. 1973. The glucose receptor. Adefective mechanism in diabetes mellitus distinct from beta adrenergicreceptor. J. Clin. Invest. 52:870-876.

5. Cerasi, E., R. Luft, and S. Efendic. 1972. Decreased sensitivity ofthe pancreatic beta cells to glucose in prediabetic and diabetic subjects.Diabetes. 21 :224-234.

6. Weir, G. C., E. T. Clore, C. J. Zmachinski, and S. Bonner-Weir.1981. Islet secretion in a new experimental model for non-insulin-de-pendent diabets. Diabetes. 30:590-595.

7. Giroix, M. H., B. Portha, M. Kergoat, D. Bailbe, and L. Picon.1983. Glucose insensitivity and amino-acid hypersensitivity of insulinrelease in rats with non-insulin-dependent diabetes. Diabetes. 32:445-451.

8. Grill, V., and M. Rundfeldt. 1986. Abnormalities of insulinresponses after ambient and previous exposure to glucose in strepto-zocin-diabetic and dexamethasone-treated rats. Diabetes. 35:44-51.

9. Leahy, J. L., S. Bonner-Weir, and G. C. Weir. 1984. Abnormalglucose regulation of insulin secretion in models of reduced (-eellmass. Diabetes. 33:667-673.

10. Deckert, T., V. Birk Lauridsen, S. Nistrup Madsen, and P.Mogensen. 1972. Insulin responses to glucose, tolbutamide, secretin,and isoprenaline in maturity-onset diabetes mellitus. Dan. Med. Bull.19:222-226.

11. Ward, W. K., J. C. Beard, J. B. Halter, M. A. Pfeiffer, and D.Porte, Jr. 1984. Pathophysiology of insulin secretion in non-insulin-dependent diabetes mellitus. Diabetes Care 7:491-502.

12. Foglia, V. G. 1944. Caracteristicas de la diabetes en la rata. Rev.Soc. Argent. Biol. 20:21-37.

13. Rossetti, L., D. Smith, G. I. Shulman, D. Papachristou, andR. A. DeFronzo. 1987. Correction of hyperglycemia with phlorizinnormalizes tissue sensitivity to insulin in diabetic rats. J. Clin. Invest.79:1510-1515.

14. DeFronzo, R. A., J. Tobin, and R. Andres. 1979. Glucoseclamp technique: a method for quantifying insulin secretion and resis-tance. Am. J. Physiol. 237:E214-E223.

15. Albano, J. D. M., R. P. Ekino, I. Maritz, and R. C. Turner.1972. A sensitive, precise radioimmunoassay of serum insulin relyingon charcoal separations of bound and free moieties. Acta Endocrinol.70:487-509.

16. Hedeskov, C. J. 1980. Mechanism of glucose-induced insulinsecretion. Physiol. Rev. 60:442-509.

17. Efendic, S., E. Cerasi, and R. Luft. 1971. Role of glucose inarginine-induced insulin release in man. Metab. Clin. Exp. 20:568-579.

18. Ward, W. K., J. B. Halter, J. C. Beard, and D. Porte, Jr. 1984.Adaptation of B and A cell function during prolonged glucose infusionin human subjects. Am. J. Physio. 246:E405-E41 1.

19. Del Prato, S., P. Sheehan, F. Leonetti, and D. Simonson. 1986.Effect of chronic physiologic hyperglycemia on insulin secretion andglucose metabolism. Diabetes. 35(Suppl. 1):748. (Abstr.)

20. Leahy, J. L., S. Bonner-Weir, and G. C. Weir. 1985. Abnormalinsulin secretion in a streptozocin model of diabetes. Diabetes.34:660-666.

21. Andrews, W. J., B. Vasquez, M. Nagulesparan, I. Klimes, J.Foley, R. Unger, and G. M. Reaven. 1984. Insulin therapy in obese,non-insulin-dependent diabetes induces improvements in insulin ac-tion and secretion that are maintain for two weeks after insulin with-drawal. Diabetes. 33:634-642.

22. Garvey, W. T., J. M. Olefsky, J. Griffin, R. F. Hamman, and0. G. Kolterman. 1985. The effect of insulin treatment on insulinsecretion and insulin action in type II diabetes mellitus. Diabetes.34:222-234.

23. Ganda, 0. P., S. Srikanta, S. J. Brink, M. A. Morris, R. E.Gleason, S. Soeldner, and G. S. Eisenbarth. 1984. Differential sensitiv-ity to ,B-cell secretagogues in "early" type I diabetes mellitus. Diabetes.33:516-521.

24. Pfeiffer, M. S., J. B. Halter, and D. Porte, Jr. 1981. Insulinsecretion of diabetes mellitus. Am. J. Med. 70:579-588.

25. Cerasi, E., R. Luft, and S. Efendic. 1972. Decreased sensitivityof the pancreatic beta cells to glucose in prediabetic and diabetic sub-jects: a glucose obese response study. Diabetes. 21:224-234.

26. Simpson, R. G., R. Benedetti, G. M. Grodsky, J. H. Karam,and P. H. Forsham. 1968. Early phase of insulin release. Diabetes.17:684-692.

27. Turner, R. C., S. T. McCarthy, R. R. Holman, and E. Harris.1976. Beta-cell function improved by supplementing basal insulin se-cretion in mild diabetes. Br. Med. J. 1: 1252-1254.

28. Vague, P., and J.-P. Moulin. 1982. The defective glucose-sensi-tivity of the (3-cell in non-insulin dependent diabetes. Improvementafter twenty hours of normoglycemia. Metab. Clin. Exp. 31:139-142.

29. Orland, M. J., R. Chyn, and M. A. Permutt. 1985. Modulationof proinsulin messenger RNAafter partial pancreatectomy in rats. J.Clin. Invest. 75:2047-2055.

30. Kaufmann, F., and R. R. Rodriguez. 1984. Subtotal pancre-atectomy in five different rat strains: incidence and course of develop-ment of diabetes. Diabetologia. 27:38-43.

31. Leahy, J. L., H. E. Cooper, and G. C. Weir. 1986. Chronichyperglycemia is associated with impaired glucose influence on insulinsecretion. J. Clin. Invest. 77:908-915.

32. Liljenquist, J. E., D. L. Horwitz, A. S. Jennings, J. L. Chiasson,U. Keller, and A. H. Rubenstein. 1978. Inhibition of insulin secretionby exogenous insulin in normal man as demonstrated by C-peptideassay. Diabetes. 27:563-570.

33. Westermark, P., and E. Wilander. 1978. The influence of amy-


loid deposits on the islet volume in maturity-onset diabetes mellitus.Diabetologia. 15:417-421.

34. Gepts, W., and P. M. Lecompte. 1981. The pancreatic islets indiabetes. Am. J. Med. 70:105-114.

35. Gjedde, A., and C. Crone. 1981. Blood-brain glucose transfer:repression in chronic hyperglycemia. Science (Wash. DC). 214:456-457.

36. Matthei, S., R. Horuk, and J. M. Olefsky. 1986. Blood-brainglucose transfer in diabetes mellitus. Decreased number of glucosetransporters at blood-brain barrier. Diabetes. 35:1081-1084.

37. DeFronzo, R. A., E. Ferrannini, and V. Koivisto. 1983. Newconcepts in the pathogenesis and treatment of non-insulin dependentdiabetes mellitus. Am. J. Med. 74(Suppl. IA):52-8 1.

38. Kolterman, 0. G., R. S. Gray, J. Griffin, P. Burstein, J. Insel,J. A. Scarlett, and J. M. Olefsky. 1981. Receptor and postreceptordefects contribute to the insulin resistance in noninsulin-dependentdiabetes mellitus. J. Clin. Invest. 68:957-969.

39. Pfeifer, M. A., J. B. Halter, and D. Porte, Jr. 1981. Insulinsecretion in diabetes mellitus. Am. J. Med. 70:579-588.

40. DeFronzo, R. A., R. Hendler, and N. Christensen. 1980. Stimu-

lation of counterregulatory hormonal responses in diabetic man by afall in glucose concentration. Diabetes. 29:125-131.

41. Henry, R. R., P. Wallace, and J. 0. Olefsky. 1986. Effects ofweight loss on mechanisms of hyperglycemia in obese non-insulindependent diabetes mellitus. Diabetes. 35:990-998.

42. Savage, P. J., L. J. Bennion, E. V. Flock, M. Nagulesparan, D.Mott, J. Roth, R. H. Unger, and P. H. Bennett. 1979. Diet-inducedimprovement of abnormalities in insulin and glucagon secretion andin insulin receptor binding in diabetes mellitus. J. Clin. Endocrinol.Metab. 48:999-1007.

43. Garvey, W. T., J. M. Olefsky, J. Griffin, R. F. Hamman, and0. G. Kolterman. 1985. The effect of insulin treatment on insulinsecretion and insulin action in type II diabetes mellitus. Diabetes.34:222-234.

44. Kolterman, 0. G., R. S. Gray, G. Shapiro, J. A. Scarlett, J.Griffin, and J. M. Olefsky. 1984. The acute and chronic effects ofsulfonylurea therapy in type II diabetes. Diabetes. 33:346-354.

45. Kosaka, K., T. Kuzuya, Y. Akanuma, and R. Hagura. 1980.Increase in insulin response after treatment of overt maturity-onsetdiabetes is independent of the mode of treatment. Diabetologia.18:23-28.


Documents

Effect Hyperglycemia Vivo Secretion Partially ... · glucose profile by phlorizin treatment in diabetic rats com-pletely corrected all three beta cell abnormalities. These re-